There’s a Clutean definition of “hard fantasy” in the Encyclopedia of Fantasy:
A useful term for stories where Magic is regarded as an almost scientific force of Nature, and subject to the same sorts of rules and principles. This was the type of fantasy championed by John W Campbell Jr in Unknown […]
That seems straightforward. Wikipedia has a compatible opinion, considering the Harold Shea stories (L. Sprague de Camp and Fletcher Pratt. I read and enjoyed these a long time ago in The Compleat Enchanter omnibus) to be the prototype of their “hard fantasy” canon: fantasy that treats magic as a predictable, rule-governed part of the natural order. Shea wins “The Roaring Trumpet” by observing, extrapolating and manipulating laws of magic—the law of contagion and the law of similarity, IIRC—if that sounds very 1940 to you, that’s because it was. The Harold Shea stories were published by Campbell in Unknown, so that’s also consistent with the previous definition.
For more recent examples of this form, the Wikipedia article also cites two of my favourite fantasy novels, both from the 90s: The Iron Dragon’s Daughter by Michael Swanwick and Metropolitan by Walter Jon Williams (the Wikipedia page for the latter claims, forlornly, that it is “arcanepunk”, which naturally is a red link of death: a sign that it’s not so easy to coin a new genre. After all, if adding -punk to stuff was enough then @everypunk has already invented twenty-five thousand new subgenres. No, genre/canon formation is a more complicated business than that, but more on that another time.) Both books treat magic in a much more complex way that the Harold Shea stories (unsurprisingly, given the half-century gap between them); both posit magic as wild, numinous and in some configurations transcendental, but nevertheless both mostly present it as a learnable skill, so I suppose it still fits. But at this point one has to ask, what isn’t hard fantasy by this definition? The same argument would do for the Chrestomanci books by Dianna Wynne Jones, the wonderful Vita Nostra by Marina and Sergey Dyachenko, and of course, Harry Potter—perhaps even more than the original, the fascinatingly nerdy Harry Potter and the Methods of Rationality fanfic by Elizier Yudkovsky, which is basically a hard(er) fantasy remake with intelligent versions of the characters. Magic is now very frequently treated as a rule-based system in popular fantasy. It seems that fantasy’s hardness has become a kind of chronic priapism since the days of Harold Shea.
Meanwhile, Jane Lindskold argues for a more rigorous definition in an 2009 essay at Tor.com: that the fantasy should aim for better “realism” in the non-fantastical parts. So magically intelligent wolves are ok; magically intelligent wolves that nod and wag their tails are not ok, because those are not wolf behaviours; magically intelligent wolves standing on their hind legs wearing natty waistcoats and inviting you a tea party, presumably, would be extremely not ok. Marie Brennan talks about another version of hard fantasy in a 2008 essay, arguing that hard fantasy, by analogy with hard science fiction, is concerned with deep rigour in some aspect of worldbuilding, such that you might say that Tolkien wrote hard linguistic fantasy or GRRM wrote hard political fantasy, and also that such stories are (again like hard sf) significantly concerned with how and why those things work.
These are similar but not identical arguments. The Lindskoldian version has a purist neti-neti quality that I find appealing (despite being terribly disinclined to actually write that sort of thing myself. I like reading it done by others, though). The Brennanian version, on the other hand, freely admits that a given story might only be rigorous in some aspects and therefore leaves so much room open that I’m not sure whether it holds up as a workable genre definition, though it can be an accurate description of Brennan’s own Midnight Never Come, for instance, which is clearly hard Elizabethan-political fantasy in her own terms.
Arguments in favour of greater rigour and “realism” are, despite their own problems, a bit more useful than just requiring magic to have rules. Harry Potter has rules, for instance, but is manifestly lacking in internal consistency and rigour because much of it is fable-like and was obviously not intended to work under hard fantasy constraints in the first place (this is why I find the Methods of Rationality fanfic fascinating; on the one hand, what it’s “fixing” was never broken, but on the other hand the changes it makes are interesting). A genre definition should able to rule Harry Potter out of hard fantasy, on the grounds that if it can’t, it’s not doing anything.
Lois Tilton’s definition of “hard fantasy” does not meet that bar, since it implicitly counts Harry Potter as hard fantasy. Tilton only requires that hard fantasy not be “soft fantasy”: i.e., the fantastical element must be unambiguous and not visibly or potentially a metaphor. But Mordor is industrialization and Aslan is Jesus and Cthulhu is probably black people or whatever? It’s not unusual for the ‘realism’ of fantastical elements to be ambiguous, is what I’m saying; ambiguity about whether something is ‘really’ happening or not is an old technique. All of Narnia is deniable because no time elapses in the wardrobe. All of the Land in Thomas Covenant is deniable, which was even an actual plot point as I recall, about Covenant not being sure if he’s hallucinating or not. In fact, if I remember right, the exact same ambiguity also applies to Swanwick’s The Iron Dragon’s Daughter, which is by all accounts otherwise an exemplar of hard fantasy.
Or here’s a different angle: in a great many fantastic fictions (and, for that matter, in journalism) the reading differs greatly depending on whether you read the fantastical element of ‘race’ as a metaphor for morality (“this character is racially predisposed to be violent”) or for the narrator’s bigotry (“this character is perceived as monstrous because the protagonist is a gigantic racist”). Have we not just softened vast swathes of formerly-hard fantasy merely by adjusting our perceptions to recognize overlooked ambiguities? Is a critical reading the ultimate boner-killer?
I do think “hard fantasy” might be an interesting idea: it has a tradition and an argument, at least, not in the Tiltonian “that which is not soft” sense but the rigour-oriented arguments summarized earlier. One could hazard that it is a sort of cousin to grimdark, perhaps, and a more distant cousin of hard sf, all of which are preoccupied with penetrating the unreal with their erect realisms.
I respect that the attempt to drink the entire firehose of sff short fiction every month (and worse, to find something to say about it all) is bound to give someone a very different perspective, and mostly one of exhaustion: what is sought is not so much a critically sound model that makes sense at a theoretical level, but a quick sorting algorithm to put stuff into different piles.
The problem, as I see it, is that the hard/soft dichotomy doesn’t work for this purpose either. They’re not balanced. Even if “hard fantasy” could theoretically be rehabilitated as per above, “soft fantasy” doesn’t seem to work at all. First because it’s impossible to pin down, as per above; second because it’s tied to a technique (hyperreal metaphors) rather than an ideological position (such as might fit into a manifesto, like mundane sf) or themes or content or style (like grimdark or cyberpunk or “hard fantasy” or “portal fantasy”), any of which might have made more sense. A scheme of categorization that files Harry Potter as “hard fantasy” and Thomas Covenant as “soft fantasy” is funny but not useful.
Some of the stories that Tilton would like to classify as “soft fantasy” might be just called magic realism, I suppose, not just for how fantastic elements are treated but for the aspect of social or political critique. But perhaps magic realism is not for white people (boom, hashtag) so this might be, as ppl say, “problematic”? How about “post-expressionist fantasy” instead? It means the same thing and sounds posh—I’m going to start using that with a straight face from now on, btw—and I also feel like genre names really need to up their game. Fantasy is being shortchanged with these monosyllables: high, low, hard, soft, &c., which people can just navigate by connotation and rumour. Whereas with post-expressionist fantasy, you’re at least forcing people to look it up, which should result in a better class of mockery if nothing else. Everybody wins.
Can’t believe I actually said all this with a minimum of dick jokes.
Today’s unrelated reading recommendation & palate cleanser: “When The Seltzer’s Like ‘Nah’ and You’re Like ‘Oh Please, Please Be Cold’ and the Seltzer’s Like ‘Nah’” by Rahawa Haile in Midnight Breakfast: “The thing about ugly is ugly’s okay so long as you’re tall. Men figured this out early and now they’re everywhere: tall, ugly men.”